Loss of microRNA-7a2 induces hypogonadotropic hypogonadism and infertility

Kashan Ahmed, … , Mathieu Latreille, Markus Stoffel

J Clin Invest. 2017;127(3):1061-1074. https://doi.org/10.1172/JCI90031.

Research Article Endocrinology Reproductive biology

MicroRNAs (miRNAs) are negative modulators of expression that fine-tune numerous biological processes. miRNA loss-of-function rarely results in highly penetrant phenotypes, but rather, influences cellular responses to physiologic and pathophysiologic stresses. Here, we have reported that a single member of the evolutionarily conserved miR-7 family, miR-7a2, is essential for normal pituitary development and hypothalamic-pituitary-gonadal (HPG) function in adulthood. Genetic deletion of mir-7a2 causes infertility, with low levels of gonadotropic and sex steroid , small testes or , impaired spermatogenesis, and lack of ovulation in male and female mice, respectively. We found that miR-7a2 is highly expressed in the pituitary, where it suppresses golgi 1 (GLG1) expression and downstream bone morphogenetic 4 (BMP4) signaling and also reduces expression of the F2a receptor negative regulator (PTGFRN), an inhibitor of prostaglandin signaling and follicle-stimulating (FSH) and (LH) secretion. Our results reveal that miR-7a2 critically regulates sexual maturation and reproductive function by interconnecting miR-7 genomic circuits that regulate FSH and LH synthesis and secretion through their effects on pituitary prostaglandin and BMP4 signaling.

Find the latest version: https://jci.me/90031/pdf The Journal of Clinical Investigation RESEARCH ARTICLE

Loss of microRNA-7a2 induces hypogonadotropic hypogonadism and infertility

Kashan Ahmed,1,2 Mary P. LaPierre,1,2 Emanuel Gasser,1,2 Rémy Denzler,1,2 Yinjie Yang,1,2 Thomas Rülicke,3 Jukka Kero,4 Mathieu Latreille,1,2 and Markus Stoffel1,2 1Institute of Molecular Health Sciences, and 2Competence Center for Systems Physiology and Metabolic Disease, Swiss Federal Institute of Technology (ETH Zurich), Zurich, Switzerland. 3Institute of Laboratory Animal Science and Biomodels Austria, Department of Biomedical Sciences, University of Veterinary Medicine Vienna, Austria. 4Departments of Pediatrics and Physiology, Institute of Biomedicine, University of Turku, and Turku University Hospital, Turku, Finland.

MicroRNAs (miRNAs) are negative modulators of that fine-tune numerous biological processes. miRNA loss-of-function rarely results in highly penetrant phenotypes, but rather, influences cellular responses to physiologic and pathophysiologic stresses. Here, we have reported that a single member of the evolutionarily conserved miR-7 family, miR- 7a2, is essential for normal pituitary development and hypothalamic-pituitary-gonadal (HPG) function in adulthood. Genetic deletion of mir-7a2 causes infertility, with low levels of gonadotropic and sex steroid hormones, small testes or ovaries, impaired spermatogenesis, and lack of ovulation in male and female mice, respectively. We found that miR-7a2 is highly expressed in the pituitary, where it suppresses golgi glycoprotein 1 (GLG1) expression and downstream bone morphogenetic protein 4 (BMP4) signaling and also reduces expression of the prostaglandin F2a receptor negative regulator (PTGFRN), an inhibitor of prostaglandin signaling and follicle-stimulating hormone (FSH) and luteinizing hormone (LH) secretion. Our results reveal that miR-7a2 critically regulates sexual maturation and reproductive function by interconnecting miR-7 genomic circuits that regulate FSH and LH synthesis and secretion through their effects on pituitary prostaglandin and BMP4 signaling.

Introduction physiology via stimulation of sex steroid synthesis (1). In males, FSH Sexual maturation and reproduction are regulated by the hypo- stimulates proliferation of immature Sertoli cells and spermatogo- thalamus, pituitary, and gonads. These endocrine organs form an nia, whereas LH stimulates Leydig cells to produce integrated system known as the hypothalamic-pituitary-gonadal (4). In females, LH triggers ovulation, promotes development of the (HPG) axis, which is responsible for the adequate secretion of male corpus luteum, and stimulates theca cells to produce androgens, and female sex hormones (1). The endocrine components of the whereas FSH stimulates recruitment of secondary ovarian follicles reproductive system are hierarchically organized and integrated in and the secretion of from granulosa cells, thereby promot- a classical endocrine feedback loop. The HPG axis is active during ing follicular maturation, which ultimately leads to ovulation (5). development and shortly after birth, silenced during childhood Lack of activation of the HPG axis leads to hypogonado- by inhibitory neurotransmitters, and reawakened at the onset of tropic hypogonadism, which is defined as low levels of the sex puberty by pulsatile secretion of gonadotropin-releasing hormone steroids testosterone or (hypogonadism) in male or (GnRH), a decapeptide that is synthesized by neurons located in the female patients, respectively, and inappropriately low levels of mediobasal (2, 3). During development, GnRH neu- gonadotropins FSH and LH (6). An increasing number of rons originate from the olfactory placode of the olfactory system and have been implicated in the molecular pathogenesis of congen- migrate to the hypothalamus, where they control reproduction by ital isolated hypogonadotropic hypogonadism (IHH), under- secreting GnRH into a capillary network that transmits GnRH to the lining the genetic and clinical heterogeneity and complexity of to stimulate secretion of follicle-stimulating hor- this condition (7–9). Genetic causes of GnRH and gonadotropin mone (FSH, encoded by Fshb) and luteinizing hormone (LH, encod- deficiency can be grouped according to the nature of their patho- ed by Lhb). Secreted GnRH acts via the GnRH receptor, which is genic mechanisms, which comprise defects in neuropeptides and expressed on gonadotropic cells in the anterior . This involved in the development and migration of GnRH action regulates synthesis and release of both gonadotropins, LH neurons (caused by mutations in KAL1, FGF8/FGFR1, PROK2/ and FSH, that control gonadal maturation and adult reproductive PROKR2, CHD7, HS6ST1, WDR11, FEZF1, NSMF and SEMA3A) in genes controlling GnRH secretion and action (GNRH1, GPR54 [also known as KISS1], TAC3/TACR3, LEP/LEPR, and GNRHR, Related Commentary: p. 796 respectively) and genes participating in synthesis of LH or FSH β subunits (7–9). Interestingly, homozygous inactivating mutations Conflict of interest: M. Stoffel is a member of the scientific advisory board of in the genes encoding the ubiquitin E3 ligase RNF216 and the Regulus Therapeutics. deubiquitinase OTUD4 have recently been identified by whole- Submitted: August 10, 2016; Accepted: December 15, 2016. Reference information: J Clin Invest. 2017;127(3):1061–1074. exome sequencing in patients exhibiting a syndrome of hypogo- https://doi.org/10.1172/JCI90031. nadotropic hypogonadism, progressive ataxia, and dementia,

jci.org Volume 127 Number 3 March 2017 1061 RESEARCH ARTICLE The Journal of Clinical Investigation

Figure 1. Ablation of mir-7a2 leads to male hypogonadism. (A–C) Representative images of mir-7a2 WT (+/+) control and KO (–/–) testes (A) and quantifica- tion of testes weights (B) of mir-7a1 KO, mir-7a2 KO, or respective WT controls (mir-7a1 control, n = 7; mir-7a1 KO, n = 6; mir-7a2 control, n = 14; mir-7a2 KO, n = 9) at 6 to 8 weeks of age or (C) at 14 days (mir-7a2 control, n = 4; mir-7a2 KO, n = 2). Scale bar: 5 mm. (D and E) Immunohistological images (D) and quanti- fication (E) of CYP17A1-positive Leydig cells (WT, mir-7a2 KO, n = 3). Scale bars: 50 μm. Arrows show Leydig cells. (F) Relative expression levels of steroidogenic genes in testes of mir-7a2 KO or control mice (WT, n = 9; mir-7a2 KO, n = 7). (G and H) Representative images of seminal vesicles (G) and quantification of weights (H) of control (upper) or mir-7a2 KO (lower) mice (WT, n = 6; mir-7a2 KO, n = 4). Scale bar: 5 mm. (I) Intratesticular testosterone levels normalized per total protein content in mir-7a2 KO or control mice (WT, n = 5; mir-7a2 KO, n = 3). (J) Total sperm count of the cauda epididymidis of 12-week-old mir-7a2 KO or control mice (WT, n = 5; mir-7a2 KO, n = 3). All data are represented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001, ANOVA (B); t test (C, E, F, H, I, J).

thereby linking disordered ubiquitination to neurodegeneration MicroRNAs (miRNAs) are small noncoding RNAs that act and reproductive dysfunction (10). Despite these recent advanc- as negative regulators of gene expression by binding to miRNA- es, the genetic basis of many cases of congenital IHH remains response elements (MREs) located primarily in 3′ UTRs of mRNA unknown, as these mutations account for approximately 50% of targets to regulate their stability and translation (12). Mammalian all IHH cases (11). Therefore, discovering additional gene muta- genomes have more than 500 miRNA genes, with miRNAs from tions will continue to advance our understanding of this syndrome. individual gene families often targeting hundreds of different

1062 jci.org Volume 127 Number 3 March 2017 The Journal of Clinical Investigation RESEARCH ARTICLE mRNAs (13, 14). With more than half of all human mRNAs esti- approximately 50% reduced testes weights at pre- and pubertal mated to be conserved miRNA targets, miRNAs are thought to ages (2 and 6 to 8 weeks old, respectively) (Figure 1, A–C). The have widespread effects on gene regulation. Many miRNA knock- body weights and lengths of mir-7a2 KO mice were comparable out models in mice show no apparent defect under normal condi- with those of WT or mir-7a1 KO animals (Supplemental Figure 1, E tions; however, they frequently exhibit severe miRNA-dependent and F). Histological analysis of mir-7a2 KO testes revealed smaller phenotypes when specific stresses are applied (15). Moreover, the but developed seminiferous tubules, with diminished numbers of importance of individual miRNAs in normal physiology and dis- spermatocytes, spermatids, and spermatozoa (Supplemental Fig- ease has been established by the discovery of mutations in miRNAs ure 1G). Immunohistochemical quantification of the Leydig cell and their targets (16–18). Recently, the collective role of miRNAs marker CYP17A1 showed that testosterone-producing Leydig cells in the HPG axis was demonstrated by genetic ablation of Dicer1 in were reduced by more than 50% in mir-7a2 KO mice (Figure 1, D GnRH neurons (19) and gonadotrophs (20), resulting in hypogo- and E). We analyzed expression levels of steroidogenic enzymes in nadism and infertility. Furthermore, miR-200b and miR-429, both testes and measured an approximately 50% reduction of Cyp17a1 members of the miR-200 family, as well as their target, ZEB1, are and Hsd3b6 mRNAs, which are required for dehydroepiandroste- required for female ovulation and reproduction in mice (21). rone and testosterone synthesis, respectively (Figure 1F). Interest- The miR-7 family is evolutionarily highly conserved and con- ingly, expression levels of the FSH receptor (Fshr) and LH receptor sidered a prototypical neuroendocrine miRNA, being expressed (Lhcgr) were increased by 2-fold, most likely due to a compensa- at high levels in neurons and neuroendocrine organs, most nota- tory response to the decreased plasma gonadotrophin levels (29). bly the endocrine , pituitary, and adrenal glands (14, 22, The sizes and weights of seminal vesicles, a classic target organ 23). Mature miR-7 harbors a conserved seed sequence that is pro- of testosterone, were reduced by 75% (Figure 1, G and H). Con- cessed from 3 precursors – miR-7a1, miR-7a2, and miR-7b – that comitantly, levels of intratesticular testosterone were decreased in are transcribed from unlinked loci on murine 13, 7, mir-7a2 KO mice (Figure 1I) and the total sperm count of the cauda and 17, respectively. A novel mechanism of miR-7 regulation was epididymidis, where mature sperm is stored, revealed a 75% recently described in neuronal cells through the identification of decrease in mir-7a2 KO mice as compared with WT controls (Fig- a brain-enriched circular RNA (known as antisense to the cerebel- ure 1J). Finally, the inguinal fat pad weights were increased 260% lar degeneration–related protein 1 transcript [Cdr1as]) comprising in 16-week-old mice (Supplemental Figure 1H). several dozens of conserved miR-7–binding sites counteracting Analogous to the phenotype observed in males, mir-7a2 KO repression by this miRNA (24, 25). Functionally, miR-7 has been females had normal body weight and length (Supplemental Fig- implicated in organ differentiation and development (26) as well ure 1, I and J), but exhibited a reduced size of estrogen hormone as in mammalian neurodegenerative diseases and cancer (27, 28). target organs, as evidenced by reduced weights and thread- Mouse genetic studies so far have revealed a role in secre- like atrophied uteri compared with WT or mir-7a1 KO mice (Fig- tion and differentiation of pancreatic β cells (22). ure 2, A and B). Histological analysis revealed that ovaries of mir- In this study, we describe the unexpected finding of a hypogo- 7a2 KO mice were able to form secondary and tertiary follicles nadotropic hypogonadism phenotype in male and female mice as a that were comparable to those of WT mice; however, they lacked result of the single mir-7a2 deletion. We identify miR-7a2 as a positive late-stage antral follicles or corpora lutea, consistent with failure regulator of FSH and LH synthesis by controlling a network involv- of folliculogenesis and ovulation (Figure 2, C and D). Expression ing bone morphogenetic protein 4 (BMP4) and prostaglandin signal- analysis of steroidogenic genes in mir-7a2 KO ovaries revealed a ing. Our data reveal the indispensability of a single miRNA for nor- drastic decrease of Cyp19a1 mRNA, also known as aromatase, the mal pituitary gonadotroph function, raising the question of whether key enzyme in estradiol biosynthesis (Figure 2E). Consequently, genetic or acquired defects leading to altered miR-7 levels may be plasma levels of estradiol were reduced by 35% in mir-7a2 KO responsible for puberty and infertility disorders of pituitary origin. female mice (Figure 2F). Furthermore, as in male mice, gonadal fat pad weights were increased and dual energy x-ray absorptiom- Results etry (DEXA) measurements of fat and lean body mass revealed Genetic ablation of mir-7a2, but not mir-7a1 or mir-7b, leads to hypo- increased fat and decreased lean mass in 16-week-old mir-7a2 KO gonadism and infertility in mice. To investigate the physiological mice compared with WT littermate mice (Supplemental Figure 1, role of miR-7, we generated mice constitutively lacking mir-7a1 K–M), consistent with the hypogonadal-induced obesity observed (referred to as mir-7a1 KO), mir-7a2 (referred to as mir-7a2 KO), in mice and humans (30, 31). Together, these data illustrate that or mir-7b (referred to as mir-7b KO) (Supplemental Figure 1, A–D; loss of mir-7a2, but not mir-7a1 or mir-7b, is sufficient to induce supplemental material available online with this article; https:// hypogonadism and infertility in mice. doi.org/10.1172/JCI90031DS1). Heterozygous breedings of all 3 miR-7a2 is the predominant miR-7 family member and is enriched mouse strains were viable, had unaltered body weight and body in the pituitary. Based on the increased testicular expression of length, and were born at normal Mendelian ratios (Supplemental Fshr and Lhcgr, which suggests decreased plasma levels of the Figure 1, E–H). Interestingly, homozygous mir-7a2 KO male and pituitary hormones FSH and LH, we hypothesized that the gonad- female mice were infertile, while homozygous mir-7a1 KO or mir- al dysfunction and infertility observed in male and female mir- 7b KO mice remained fertile, with unaltered reproductive rates 7a2 KO mice resulted from impaired pituitary function. First, we (Supplemental Table 1). systematically analyzed expression levels of miR-7a and miR- Homozygous mir-7a2 KO male mice failed to undergo sexual 7b by absolute quantification in tissues of the hypothalamic- maturation and exhibited microphallus and hypogonadism with pituitary-adrenal/gonadal axis. We found that miR-7a and miR-7b

jci.org Volume 127 Number 3 March 2017 1063 RESEARCH ARTICLE The Journal of Clinical Investigation

Figure 2. Ablation of mir-7a2 leads to female hypogonadism and anovu- lation. (A and B) Representative images of ovaries and uteri (A) and quan- tification of ovary weights (B) of mir-7a1 KO, mir-7a2 KO, or respective control mice (mir-7a1 control, n = 9; mir-7a1 KO, n = 10; mir-7a2 control, n = 7; mir-7a2 KO, n = 4). Scale bar: 5 mm. (C and D) Histological exam- ination using H&E staining of ovaries of WT control (C) or mir-7a2 KO (D) mice (n = 4). CL, corpus luteum; LAF, large antral follicle; GF, growing follicle. Shown are representative images of 4 mice per genotype. Scale bar: 100 μm. (E) Relative expression levels of ovarian steroidogenic genes in mir-7a2 KO or control mice (WT, n = 6; mir-7a2 KO, n = 4). (F) Plasma estradiol levels of 8-week-old mir-7a2 KO or control mice (WT, n = 12; mir-7a2 KO, n = 11). All data are represented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001, ANOVA (B); t test (E, F).

were enriched in pituitary as compared with hypothalamus (10- fold less), (50-fold less), or gonads (up to 500-fold less) and that miR-7a was 2.5-fold more abundant than miR-7b in pituitary (Supplemental Figure 2, A and B). Given the low expres- sion of miR-7a in male and female gonads, it is unlikely that miR- 7a would repress target genes and exert a biological function in testis and ovary (32). As pre-miR-7a1 and pre-miR-7a2 are processed into an iden- tical mature miR-7a sequence, the TaqMan probe of the miRNA assay cannot distinguish between the 2 miRNA gene products and thus measures combined miR-7a1 and miR-7a2 levels. To assess the contribution of miR-7a1 and miR-7a2 to the total pool of miR-7a in the pituitary, we analyzed expression levels of miR- 7a in mice lacking either mir-7a1 or mir-7a2. Interestingly, total miR-7a expression was reduced by approximately 80% in mir-7a2– deficient mice, while miR-7a levels appeared to be unchanged in mice lacking mir-7a1 when compared with WT littermate animals (Supplemental Figure 2, C–F). Expression of miR-7b was unchanged in mir-7a1– and mir-7a2–deficient mice compared with WT animals, thereby confirming the specificity of the TaqMan assay. In conclusion, miR-7a2 is highly enriched in pituitary com- pared with other neuroendocrine tissues; furthermore, it accounts for the large majority of total pituitary miR-7 copies, and miR-7a1 and miR-7b cannot compensate for loss of miR-7a2 expression. miR-7a2 ablation causes hypopituitarism and hypogonadotrop- ic hypogonadism. To test whether the hypogonadism observed in mir-7a2 KO mice was caused by altered gene expression of essential pituitary genes, we performed RNA sequencing (RNA- Seq) of pituitaries from 6-week-old WT and mir-7a2 KO mice. We measured a dramatic decrease of gonadotropin expression with a reduction of Fshb mRNA by 86% (FDR = 5.94 × 10-52) and Lhb mRNA by 78% (FDR = 1.26 × 10-82) in mir-7a2–deficient pituitaries that was already apparent in 2-week-old mice (Figure 3A and Sup- plemental Figure 3A). In addition, expression of the glycoprotein hormone α subunit (Cga) and (Prl) mRNAs was reduced, while the expression of the GnRH receptor (Gnrhr) and other hor- mones of the anterior pituitary, including stimulating hor- mone β (Tshb), (Gh), and pro-opiomelanocortin -α (Pomc), remained unchanged. Since lack of Prl does not lead to hypogonadism (33), we focused our investigation on the charac- terization of gonadotroph cells. Immunohistochemical analysis of pituitary sections revealed reduced numbers of FSH-expressing (–66%) and LH-expressing (–60%) cells in mir-7a2 KO mice (Fig- ure 3, B–E). We also observed that pituitary weights were reduced

1064 jci.org Volume 127 Number 3 March 2017 The Journal of Clinical Investigation RESEARCH ARTICLE

Figure 3. Constitutive genetic ablation of mir-7a2 causes hypogonadotropic hypogonadism. (A) Expression of pituitary hormones in male mir-7a2 KO or control mice shown as heat map analysis from RNA-Seq (WT, mir-7a2 KO, n = 3). (B and C) Representative immunohistological images of pituitary sections (n = 3) stained for LH (B) or FSH (C) of control (upper images) or mir-7a2 KO mice (lower images). Scale bars: 200 μm. (D and E) Quantification of LH-positive (D) or FSH-positive (E) cells in pituitary of mir-7a2 KO or control mice (WT, mir-7a2 KO, n = 3). (F and G) Pituitary weights of male (F) or female (G) mir-7a1 KO, mir-7a2 KO, or respective control mice (males, mir-7a1 control, mir-7a1 KO, n = 4; mir-7a2 control, mir-7a2 KO, n = 8; mir-7a1 control, n = 8; mir-7a1 KO, n = 10; mir-7a2 control, n = 6; mir-7a2 KO, n = 7). (H and I) Plasma levels of FSH (H) or LH (I) in male mir-7a2 KO or control mice (WT, mir-7a2 KO, n = 7). (J and K) Plasma levels of FSH (J) or LH (K) in female mir-7a2 KO or control mice (WT, mir-7a2 KO, n = 4). (L) Number of oocytes collected after superovulation test in 5-week-old mir-7a2 KO or control mice (WT, n = 4; mir-7a2 KO, n = 3). All data are represented as mean ± SD except in (H, K), where data are represented as ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001, ANOVA (F, G); t test (D, E, H, I, J, K).

in both male and female mir-7a2 KO mice (Figure 3, F and G). impaired and that the defect in mir-7a2 KO mice is primarily at the Importantly, the observed alterations of gene expression resulted level of the pituitary (Supplemental Figure 3, B and C). We then in strongly decreased plasma concentrations of FSH and LH in tested whether the hypopituitarism observed in mir-7a2 KO mice male and female mir-7a2 KO mice (Figure 3, H–K). Furthermore, was responsible for the impaired gonadal function and anovu- a GnRH stimulation test with buserelin revealed an increase of lation by subjecting female mice to a standard superovulation plasma FSH and LH levels in WT, but not mir-7a2 KO mice, fur- treatment using pregnant mare serum. Ovulation was induced in ther confirming that the response of gonadotropic cells to GnRH is hormone-treated mir-7a2 KO mice, indicating that ovaries were

jci.org Volume 127 Number 3 March 2017 1065 RESEARCH ARTICLE The Journal of Clinical Investigation

Figure 4. Acute ablation of mir-7a2 in adult mice recapitulates hypogonadotropic hypogonadism. (A) TaqMan assays showing relative expression levels of miR-7a2, miR-7b, or miR-16 in pituitary of male UBC-Cre × mir-7a2fl/fl or Cre-negative mir-7a2fl/fl mice 10 weeks after a 5-day treatment with tamoxifen (TAM) or vehicle (UBC-Cre × mir-7a2fl/fl + vehicle, n = 5, mir-7a2fl/fl + TAM, UBC-Cre × mir-7a2fl/fl + TAM, n = 4). (B) Relative expression levels of pituitary hor- mones in UBC-Cre × mir-7a2fl/fl or mir-7a2fl/fl mice after treatment with tamoxifen or vehicle (UBC-Cre × mir-7a2fl/fl + vehicle, n = 5, mir-7a2fl/fl + TAM, UBC-Cre × mir-7a2flox + TAM, n = 4). (C and D) Immunohistological quantification of LH-positive (C) or FSH-positive (D) cells in pituitary sections of UBC-Cre × mir-7a2fl/fl or mir-7a2fl/fl mice after treatment with tamoxifen (mir-7a2fl/fl + TAM, n = 4; UBC-Cre × mir-7a2fl/fl + TAM, n = 3). (E and F) Plasma levels of FSH (E) or LH (F) of male UBC-Cre × mir-7a2fl/fl or mir-7a2fl/fl mice after treatment with tamoxifen or vehicle (FSH, UBC-Cre × mir-7a2fl/fl + vehicle, n = 5; mir-7a2fl/fl + TAM, UBC-Cre × mir-7a2fl/fl + TAM, n = 4; LH, UBC-Cre × mir-7a2fl/fl + vehicle, mir-7a2fl/fl + TAM, UBC-Cre × mir-7a2fl/fl + TAM, n = 3). (G) Testes weights of UBC-Cre × mir-7a2fl/fl or mir-7a2fl/fl mice after treatment with tamoxifen or vehicle (UBC-Cre × mir-7a2fl/fl + vehicle, n = 6; mir-7a2fl/fl + TAM, n = 10; UBC-Cre × mir-7a2fl/fl + TAM, n = 8). (H) Relative expression levels of the Leydig cell marker CYP17A1 in testes of UBC-Cre × mir-7a2fl/fl or mir-7a2fl/fl mice after treatment with tamoxifen or vehicle (UBC-Cre × mir-7a2fl/fl + vehicle, mir-7a2fl/fl + TAM, UBC-Cre × mir-7a2fl/fl + TAM, n = 4). All data are repre- sented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001, ANOVA (A, B, E, F, G, H); t test (C and D).

principally functional and that hypophyseal-pituitary dysfunction trolling GnRH secretion and action were similar in hypothalami was the primary cause for the hypogonadism in mir-7a2–ablated of mir-7a2 KO and control littermates (Supplemental Figure 3, G mice (Figure 3L). We can exclude that the relatively low expres- and H). Interestingly, the neuropeptides 1 (Kiss1) and sion of miR-7a in the hypothalamus affects GnRH-expressing tachykinin 2 (Tac2) were significantly increased in hypothalami neurons, as Gnrh mRNA levels, morphology, and expected pro- of mir-7a2 KO mice, which corroborates findings that Kiss1 and jections of GnRH-expressing neurons were unaltered in mir-7a2 Tac2 expression are under inhibitory modulation from circulating KO mice (Supplemental Figure 3, D–F). Furthermore, transcript gonadal hormones (34, 35). In addition, olfactory bulbs appeared levels of neuropeptides and genes involved in the development to have normal size and architecture and no difference was detect- and migration of GnRH neurons and expression of genes con- ed in behavioral olfactory testing of male and female mice lacking

1066 jci.org Volume 127 Number 3 March 2017 The Journal of Clinical Investigation RESEARCH ARTICLE

Figure 5. Pituitary gene expression analysis and identification of target genes of mir-7a2. (A) Expression of key pituitary markers in mir-7a2 KO or control mice shown as heat map analysis from RNA-Seq of 6-week-old male mice (WT, mir-7a2 KO, n = 3). (B) Relative expression of key pituitary markers in UBC- Cre × mir-7a2fl/fl or Cre-negative mir-7a2fl/fl mice 10 weeks after a 5-day treatment with tamoxifen or vehicle (UBC-Cre × mir-7a2fl/fl + vehicle, n = 5, mir-7a2fl/fl + TAM, UBC-Cre × mir-7a2fl/fl +TAM, n = 4). (C) RNA-Seq results from pituitary of 6-week-old male mir-7a2 KO or control mice. Cumulative distributions of mRNA changes for predicted target genes of miR-7 (left)or miR-16 (right) with the indicated context+ score bins (color) or for genes with no respective miRNA site (black), as scored by the context+ model of TargetScan 6.2 (60). Number of genes per bin for miR-7: black, 11,884; green, 1,927; orange, 782; red, 245; purple, 83; miR-16: black, 12,841; green, 656; orange, 967; red, 306; purple, 151. (W T, mir-7a2 KO, n = 3). (D) Relative expression of predicted miR-7 tar- get genes that were more than 1.3-fold upregulated in RNA-Seq in pituitaries of UBC-Cre × mir-7a2fl/fl or mir-7a2fl/fl mice 10 weeks after a 5-day treatment with tamoxifen relative to vehicle-treated mice (set as 1, black dotted line) (mir-7a2fl/fl + TAM, n = 4; UBC-Cre × mir-7a2fl/fl + TAM, n = 3). (E) Relative mRNA expression levels of miR-7 targets Glg1 and Ptgfrn in pituitary of WT control (mir-7a2+/+) or mir-7a2 KO (mir-7a2–/–) mice (n = 4). All data are represented as mean ± SD. *P < 0.05; **P < 0.01, ***P < 0.001, ****P < 0.0001, ANOVA (B and C); t test (A, D, E).

mir-7a2 compared with littermate controls (data not shown, Sup- Acute genetic ablation of miR-7a2 in adult mice recapitulates plemental Figure 3I), suggesting that the hypogonadotropic phe- hypogonadotropic hypogonadism. To circumvent potential devel- notype was not due to defective GnRH neuron migration from the opmental effects caused by constitutive mir-7a2 ablation, we olfactory placode into the correct hypothalamic location during bred mice carrying a floxed mir-7a2 allele (referred to as mir- development. Collectively, these data demonstrate that mir-7a2 7a2fl/fl) with transgenic mice expressing Cre-recombinase fused deficiency causes pituitary dysfunction and mirrors the disease to a mutant form of the estrogen receptor ERT2 under the tran- pathology of human hypogonadotropic hypogonadism. scriptional control of the human ubiquitin C promoter (UBC-

jci.org Volume 127 Number 3 March 2017 1067 RESEARCH ARTICLE The Journal of Clinical Investigation

Figure 6. miR-7a2 regulates gonadotropin production through BMP and prostaglandin signaling. (A and B) Relative luciferase levels of plasmids car- rying WT or mutated 3′ UTRs of Ptgfrn (A) or Glg1 (B) cotransfected in LbT2 cells with or without forced expression of mir-7a2 (n = 3). (C and D) Relative expression levels of gonadotropin genes in cells transfected with siRNA against Ptgfrn (C) or Glg1 (D) (n = 4). (E) Relative expression levels of gonadotropin genes, Cga, Fshb, and Lhb in cells overexpressing Ptgfrn (gray bars) or Glg1 (black bars) for 72 hours (n = 4). (F and G) Relative expression levels of Fshb (F) or Lhb (G) in LbT2 cells that were transfected with siPtgfrn or siCtrl and treated with 100 nM dinoprost or PBS for 4 hours (n = 4). (H) Concentration of LH in supernatants of cells 72 hours after silencing of Ptgfrn (n = 3). (I) Expression levels of Lhb in cells treated with dinoprost (100 nM), GREM1 (0.25 μg/ml), or dinoprost and GREM1 together (n = 3). (J) Western blot analysis of phospho-SMAD1/5/9 or total SMAD1 in lysates of cells pretreated with or without GREM1 (0.25 ug/ml) for 1 hour, followed by 30 minutes of stimulation with BMP4 (50 ng/ml). Shown is 1 representative experiment of 3. All data are repre- sented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001, ANOVA (E, F, G, I); t test (A, B, C, D, H).

Cre) (36) to allow tamoxifen-inducible global ablation of mir-7a2 that received only vehicle (Figure 4B). Immunohistochemical (referred to as UBC-Cre × mir-7a2fl/fl). We treated 6-week-old quantification revealed that FSH cell area was reduced by 55% in male UBC-Cre × mir-7a2fl/fl or mir-7a2fl/fl mice with tamoxifen or tamoxifen-treated UBC-Cre × mir-7a2fl/fl mice, while LH cell area vehicle and confirmed ablation of mir-7a2 using a TaqMan miR- decreased to a lesser extent, with borderline significance (Figure NA assay 8 weeks after treatment (Figure 4A). While pituitary 4, C and D). Most notably, acute ablation of mir-7a2 in adult mice weight and body weight remained unchanged between groups led to a drastic reduction of plasma FSH and LH levels (Figure 4, E (Supplemental Figure 3, J and K), a striking 50% reduction of and F), and importantly, these hormones were unaltered between Fshb and Lhb mRNA expression levels was observed in the pitu- vehicle-treated UBC-Cre × mir-7a2fl/fl and mir-7a2fl/f mice that itaries of tamoxifen-treated UBC-Cre × mir-7a2fl/fl mice as com- received tamoxifen (Figure 4, E and F). The decreased gonadotro- pared with control mir-7a2fl/fl mice or UBC-Cre × mir-7a2fl/fl mice pin expression and secretion resulted in reduced testes weights in

1068 jci.org Volume 127 Number 3 March 2017 The Journal of Clinical Investigation RESEARCH ARTICLE

further validated by quantitative PCR (qPCR) that the mRNA tran- scripts of Ptgfrn and golgi glycoprotein 1 (Glg1), a member of the cysteine-rich fibroblast family (40), were increased in the pituitaries of mir-7a2 KO mice (Figure 5E). To test whether predicted mir-7 targets were repressed by increased miR-7 levels, we overexpressed miR-7a2 using a recombinant ade- novirus (Ad–miR-7a2) in 2 gonadotroph cell lines, aT3 and LbT2, and observed significant repression of several predicted miR-7 tar- get candidates (Supplemental Figure 4, C and D). Together, these results reveal a tissue-specific regulation of direct and secondary Figure 7. Model illustrating the molecular pathways by which miR-7a miR-7 targets in the pituitary, with strong effects on upstream regu- regulates gonadotropic hormone secretion in the pituitary. lators of prostaglandin and BMP4 signaling. miR-7 targets Ptgfrn and Glg1 regulate expression and secretion of pituitary gonadotropins. To investigate the role of the identified tamoxifen-treated UBC-Cre × mir-7a2fl/fl mice as compared with miR-7 target genes in pituitary function, we undertook an unbi- control mir-7a2fl/fl mice or UBC-Cre × mir-7a2fl/fl mice that received ased approach and silenced the 55 most regulated miR-7 targets only vehicle (Figure 4G). Moreover, testicular expression of the using smart-pool siRNAs in the aT3 and LbT2 pituitary cell lines Leydig cell marker CYP17A1 was reduced by 35% (Figure 4H). and analyzed the effect on expression levels of Fshb and Cga. We Together, these data demonstrate that acute ablation of mir-7a2 in found that silencing of both Ptgfrn and Glg1 induced hormone adult mice is sufficient to recapitulate hypopituitarism and hypo- transcription, indicating that these genes may act as negative reg- gonadism observed in constitutive mir-7a2 KO mice. ulators of pituitary hormone transcription (Supplemental Figure Dysregulated pituitary gene expression and miR-7a target dere- 5, A and B). The 3′ UTRs of Ptgfrn and Glg1 contain 1 or 2 7mer pression in mir-7a2 ablated mice. To gain insights into the molecu- miRNA-recognition sequences for miR-7, respectively, suggesting lar mechanisms of miR-7a2–dependent pituitary dysfunction, we they are direct targets of miR-7. To confirm this, we cloned the performed RNA-Seq of mir-7a2 KO and WT littermate mice and WT or mutated 3′ UTRs of Ptgfrn and Glg1 into luciferase reporter analyzed the expression of genes known to be involved in pituitary constructs, performed luciferase assays, and demonstrated that development, gonadotrope function, and/or in LH or FSH subunit both Ptgfrn and Glg1 were direct target genes of miR-7a (Figure 6, gene expression. Established key factors of early pituitary function A and B). Next, we validated the findings from the siRNA screen in (Egr1, Gnrhr, Acvr1, Acvr1b, Acvr1c, Acvr2a, Acvr2b, Smad1, Smad2, a separate experiment and detected a more than 1.5-fold upregula- Smad4–7­­ , Smad9, and Foxl2) (Supplemental Figure 4A) and devel- tion of Fshb when Ptgfrn or Glg1 was silenced by RNA interference opment (Gli2, Gli3, Nr5a1, Pitx1, Pitx2, Isl1, Tbx19, Pou1f1, Prop1, (Figure 6, C and D). In addition, knockdown of Ptgfrn increased Hes1) remained unchanged between constitutive mir-7a2 KO and Lhb expression by 1.4-fold in LbT2 cells, while Cga was modestly control mice, while Grem1, coding for gremlin 1, a secreted BMP4 regulated (Figure 6C). In contrast, overexpression of Ptgfrn or Glg1 antagonist (37), the zinc finger protein Gli1, and the LIM homeo- downregulated expression of Fshb, Lhb, and Cga (Figure 6E). box protein 4 (Lhx4) were strongly reduced in mir-7a2 KO mice PTGFRN is known to antagonize the physiological action of (Figure 5A). In tamoxifen-induced UBC-Cre × mir-7a2fl/fl mice, we prostaglandin F2 α (PGF2α), pharmaceutically also termed dino- confirmed that mRNA levels of Grem1, Gli1, and Lhx4 were similar- prost, on its cognate receptor, PTGFR (38, 39). Activation of PTG- ly reduced, suggesting that these genes may be relevant for miR- FR by dinoprost is coupled to stimulatory Gq/G11-type G proteins, 7a2–dependent effects in the pituitary (Figure 5B). Direct target which leads to an increase in intracellular calcium levels (41). To genes of miRNAs were expected to be upregulated in miRNA loss- investigate a potential mechanistic role of PGF2α signaling in gonad- of-function models. To identify potential direct targets of miR-7, otroph cells, we tested whether activation of PTGFR would induce we performed bioinformatic analysis and found that mRNAs car- the expression of pituitary hormones in LbT2 cells following treat- rying a miR-7 response element were preferentially upregulated in ment with 10 nM dinoprost. Indeed, administration of dinoprost mir-7a2 KO mice (Figure 5C). No seed enrichment was observed resulted in increased expression levels of both gonadotropin genes, for predicted targets of ubiquitously expressed mir-16, which is Fshb and Lhb, after 4 and 8 hours (Supplemental Figure 6, A and B). also abundant in the pituitary (Figure 5C). Interestingly, we did We then tested whether reduced activity of PTGFRN, the negative not measure a significant upregulation (except Pfn2 and Prkcb) regulator of PTGFR, would further enhance the observed pharma- of the strongest miR-7 targets previously described in pancreatic cological effects of PGF2α on pituitary hormone expression. To this islets to regulate insulin secretion (22) (Supplemental Figure 4B), end, we silenced Ptgfrn in gonadotroph cells and treated them with indicating a tissue-specific regulation of miR-7 target genes. Using dinoprost for 8 hours. We found that loss of Ptgfrn further amplified a cut-off of 1.3-fold upregulation, RNA-Seq revealed 55 potential the dinoprost-dependent induction of Fshb and Lhb (Figure 6, F and direct target transcripts of miR-7. We analyzed expression of these G), underscoring the specificity of a PGF2α/PTGFRN/PTGFR axis genes in tamoxifen-induced UBC-Cre × mir-7a2fl/fl mice and con- to regulate hormone expression. In addition, silencing of Ptgfrn firmed derepression of several predicted miR-7 targets (Figure 5D). resulted in increased secretion of LH in LbT2 cells (Figure 6H). Notably, Ptgfrn, the negative regulator of prostaglandin receptor In addition to the negative regulation of gonadotropin produc- F2a (38, 39), was as strongly upregulated as the established miR- tion by PTGFRN, BMP signaling has also been shown to interfere 7 target Cplx2 (Ptgfrn, 1.41-fold; Cplx2, 1.32-fold) (Figure 5D). We with hormone production in pituitary gonadotrophs (42, 43). Given

jci.org Volume 127 Number 3 March 2017 1069 RESEARCH ARTICLE The Journal of Clinical Investigation that the BMP4 antagonist Grem1 was strongly reduced in miR-7a2 termed anosmia, due to defective migration of GnRH neurons KO mice, we first tested to determine whether Grem1 overexpres- from the olfactory placode into the hypothalamus during devel- sion would increase gonadotroph expression and detected a mod- opment and the subsequent lack of GnRH release to the pituitary, est, but significant, increase in Fshb, Lhb, and Cga mRNA levels leading to hypopituitarism (47). Our results show that the hypogo- (Supplemental Figure 6C). In addition, we treated cells exogenous- nadotropic hypogonadism in mir-7a2 KO mice is highly unlikely ly with recombinant GREM1 and analyzed its effect on hormone to be causally linked to miR-7 function in hypothalamic neurons production. Interestingly, we detected an induction of hormone because (a) absolute quantification of miR-7 levels demonstrates transcription that was comparable to that of dinoprost (Figure 6I). that miR-7a expression is 10-fold lower in hypothalamus as com- Most strikingly, when dinoprost and GREM1 were coadministered, pared with pituitary and therefore, based on miR-7b levels, in a we measured an additional upregulation of hormone expression range where no phenotype and target gene regulation is expected; (Figure 6I). Mechanistically, GREM1 inhibited BMP4-induced (b) hypothalamic GnRH expression levels are indistinguishable phosphorylation of SMAD1 in LbT2 cells when these cells were between mir-7a2 KO and control littermates; (c) mir-7a2 KO mice pretreated with recombinant GREM1 (Figure 6J). Moreover, we show unaltered olfaction and normal expression of hypothalamic observed that GREM1 was upregulated in cells depleted in GLG1, genes that are known to be involved in hypothalamus-dependent while being decreased in cells overexpressing GLG1, indicating that hypopituitarism; and (d) expression of the GnRH receptor in the the direct target GLG1 was necessary for repression of GREM1 lev- pituitary, which is downregulated in animal models of GnRH defi- els (Supplemental Figure 6, D and E). Together, these data identify ciency, is expressed at similar levels in mir-7a2 KO and WT control Ptgfrn and Glg1 as direct targets of miR-7 that negatively regulate mice (48). In addition, miR-7 is not among the 53 miRNAs that are pituitary hormone transcription and secretion through regulation enriched in GnRH neurons compared with the surrounding non- of both prostaglandin- and BMP-dependent signaling pathways. GnRH cells (19), further suggesting that miR-7 is unlikely to have a specific function in these neurons. Discussion We have previously shown that loss of miR-7a2 in pancreatic The results presented in this study highlight the role of miR-7a2 in β cells leads to increased β cell function and enhanced glucose- the regulation of pituitary function and the development of hypo- stimulated insulin secretion (22). Interestingly, ablation of mir-7a2 gonadotropic hypogonadism and infertility. Our data demonstrate in the pituitary leads to impaired gonadotroph cell function and that miR-7a2 is the predominant member of the miR-7 family in hormone secretion. These opposing phenotypes, elicited by a loss the pituitary and that constitutive genetic ablation of mir-7a2, but of the same miRNA in 2 different neuroendocrine cell types, sug- not mir-7a1, is sufficient to induce hypopituitarism and infertility. gest that miR-7a regulates different networks in a tissue-specific Deletion of mir-7a2 leads to decreased expression and profoundly and context-dependent manner. The molecular mechanism for reduced plasma levels of the pituitary gonadotropins FSH and LH, this observation is currently elusive, but will likely be due to dif- resulting in hypogonadism in both male and female mir-7a2 KO ferent miRNA and target abundances and synergistic regulation of mice, as indicated by decreased steroidogenic gene expression and target genes by differentially expressed miRNAs as well as diverse reduced sex steroid levels and gonadal sizes. We also measured occupancy of transcripts by RNA-binding proteins, which may decreased expression levels of prolactin expression in constitu- influence binding of miRNAs to their target transcripts, thereby tive and inducible mir-7a2 KO mice. This finding warrants further regulating the susceptibility of miRNA regulation. investigation and suggests that miR-7a2 either is also essential for The miR-7 family has previously been implicated in organ dif- lactotroph cell function in the pituitary or represses the function of ferentiation and development (22). Our results showing already dopaminergic neurons in the arcuate nucleus of the hypothalamus reduced gonad sizes and dysregulated expression of FSH and LH that are known to inhibit prolactin secretion (44). Interestingly, as well as direct and indirect miR-7 targets in prepubertal mice aged male and female mir-7a2 KO mice have increased fat mass, a indicate that a developmental defect contributes to the hypogo- phenotype that may be caused by a long-lasting and physiological- nadotrohic phenotype observed in mice with constitutive genetic ly relevant depletion of gonadal steroid hormones (45). However, ablation of mir-7a2. However, our data obtained from temporally we cannot exclude that miR-7a2 also influences body weight and induced mir-7a2 ablation in postdevelopmental adult mice provide energy expenditure in a partially sex steroid–independent man- genetic evidence that impaired gonadotroph cell functions cannot ner, for example, through neurons in the hypothalamus, where solely be ascribed to a developmental defect and demonstrate a miR-7a2 expression has also been reported (46). The observed direct role of miR-7a in the maintenance of pituitary gonadotroph hypogonadism stems from a dysfunction of upstream components function in adult animals. Mechanistically, we identify the nega- of the HPG axis, since ovarian function and ovulation are func- tive regulator of PGF2α receptor, Ptgfrn, and Glg1 as relevant direct tionally rescued when pituitary hormones are replaced in miR-7a2 target genes of miR-7a that negatively regulate gonadotropin pro- KO mice in a superovulation test. This phenotype is reminiscent duction and secretion. While only these targets could be linked to of the disease pathology of human hypogonadotropic hypogonad- gonadotropin regulation in our pituitary cell models, we cannot ism, prompting the question of whether some cases of congential rule out that other miR-7 targets also contribute to the defect in IHH of pituitary origin with as-yet-unknown genetic causes may gonadotroph development and function of mir-7a2 KO mice. The be attributed to abnormal miR-7a2 levels. Further investigations physiological effects of on pituitary gonadotrophs are warranted to explore this possibility. have not been fully elucidated and have been in part discussed con- Cases of Kallmann’s syndrome, a form of human hypogonad- troversially. While it is accepted that prostaglandins, particularly otropic hypogonadism, are often accompanied by a loss of smell, PGE1, act on the anterior pituitary and can induce the formation of

1070 jci.org Volume 127 Number 3 March 2017 The Journal of Clinical Investigation RESEARCH ARTICLE

cyclic AMP, a direct effect on LH and FSH could not be established targeting vector was electroporated in C5BL/6N embryonic stem cells. (49). In the case of PGF2α, known as dinoprost, however, existing Upon selection with G418 (Invitrogen), clones were picked, expanded, data suggest that PGF2α treatment may regulate LH stores and and verified for integration by Southern blotting. Positive clones were release (50–52). Our data now demonstrate that in pituitary gonad- microinjected in BALB/c blastocytes and subsequently transferred otroph cell lines, PGF2α/dinoprost increases gonadotropin produc- into pseudopregnant females to generate chimeric offspring. Chime- tion and that the direct miR-7a target PTGFRN acts as a negative ras were then bred with C57BL/6 females to produce heterozygotes, modulator in a PGF2α/PTGFRN/PTGFR regulatory circuit. which were crossed with Flipase mice to delete the neor cassette. Mice In addition, we here describe the inhibitory role of the miR-7a were bred with B6.C-Tg(CMV-cre)1Cgn/j mice (Jackson Laboratory) target GLG1 in gonadotroph cell functions. While the precise phys- for deleting mir-7b. Genotyping of mir-7b KO mice was performed by iological functions of GLG1 in pituitary hormone homeostasis are PCR using primers listed in Supplemental Table 2. unknown, our data suggest that GLG1 negatively affects expres- DEXA. DEXA measurements of fat and lean body mass were per- sion of the BMP antagonist GREM1. BMP4 plays a crucial role in formed as whole animal scans using an animal CT-Scanner (LaTheta) the formation of the anterior pituitary by regulating diverse cel- at 1-mm intervals from the head to the base of the tail. Images were lular responses, such as cell differentiation, migration, adhesion, analyzed using the LaTheta software. and proliferation (53), but also has been described as inhibiting Immunostainings. Mice were perfused with 4% paraformaldehyde, FSH production (42, 54). BMPs activate BMP receptors, which in and organs were postfixed overnight at 4°C and embedded in paraffin. turn recruit and phosphorylate SMAD proteins that translocate to Microtome sections were subjected to antigen retrieval in a pressure the nucleus to repress Fshb subunit gene expression. Furthermore, oven in 10 mM sodium citrate, pH 6.0, for 10 minutes at 95°C. Sec- BMPs can antagonize the activin-mediated effects known for stim- tions were permeabilized and blocked in PBS containing 0.1% Triton ulating FSH release (42). We propose that in miR-7a2 KO mice, the X-100, 1% BSA, and 5% goat serum. Primary antibody binding was drastic reduction in GREM1 levels is partly mediated by dysinhi- performed overnight at 4°C, while secondary antibody incubation was bition of its negative regulator and miR-7 target GLG1, leading to carried out at room temperature for 1 hour. derepression of BMP4 and, consequently, increased BMP activity, For morphometric analysis of pituitaries, 5-μm sections were which affects pituitary development defect and impairs gonad- cut and 5 sections from each pituitary were stained with anti-LH or otroph function in adult mice (Figure 7). Indeed, in this context, anti-FSH antibodies and Hoechst dye to stain nuclei. Sections were we demonstrate that GREM1 acts as a BMP4 antagonist to reduce scanned using a 20× objective of a Zeiss Apotome microscope. The phosphorylation of SMAD proteins in pituitary LbT2 cells. number of FSH- or LH-positive cells compared with total cell num- In summary, our data reveal miR-7a2 as a crucial regulator of ber was determined using CellProfiler software. For morphometrical pituitary gonadotroph cell function and describe how the loss of quantification of testicular Leydig cells, 4-μm paraffin sections were mir-7a2 leads to hypogonadotropic hypogonadism. Mutations in cut and 5 sections per testis were stained with anti-CYP17A1 anti- the miR-7a2 gene and within the binding sites in the 3′ UTRs of body and Hoechst dye to stain nuclei. Sections were scanned using a its most regulated targets should be considered as potential genet- 20× objective of the Panoramic 250 Slide Scanner (3D Histech). The ic causes for hypogonadotropic hypogonadism in humans. Our area fractions of CYP17A1-positive cells relative to total cell area were findings also characterize PTGFRN and GLG1 as two novel target determined using ImageJ (Fiji, NIH) software. genes of miR-7 that negatively regulate gonadotropin production. Western blots. Cells were lysed in RIPA buffer containing protease Pharmacological stimulation of miR-7 activity and/or direct inhi- inhibitors (Roche) and HALT phosphatase inhibitors (Pierce). Protein bition of its targets PTGFRN and GLG1 may provide a basis to concentrations were measured by the Bradford method. Proteins were develop strategies to improve pituitary function in human disease. separated by SDS-PAGE, transferred to nitrocellulose membranes, and blocked in 5% milk/TBS-T or 5% BSA/TBS-T for 1 hour. Mem- Methods branes were incubated with appropriate antibodies overnight at 4°C. Animal husbandry and mouse strains. All mice were on a pure C57BL/6N Membranes were exposed to secondary antibodies for 1 hour at room background. Mice were housed in a pathogen-free animal facility at temperature and developed using ECL Western Blotting Substrate. the Institute of Molecular Health Sciences at ETH Zurich. The animals Illumina RNA-Seq and analysis. Library preparation was as follows: were maintained in a temperature- and humidity-controlled room on the quality of the isolated RNA was determined with a Qubit (1.0) a 12-hour light/12-hour dark cycle (lights on from 6 am to 6 pm). Mice Fluorometer (Life Technologies) and a Bioanalyzer 2100 (Agilent). were fed a standard laboratory chow diet. Generation of mir-7a1 KO Only those samples with a 260/280 nm ratio between 1.8 and 2.1 and and mir-7a2 KO mice was described previously (22). UBC-Cre mice a 28S/18S ratio within 1.5 and 2.0 were further processed. The TruSeq (Tg[UBC-cre/ERT2]1Ejb) were purchased from Jackson Laboratories. RNA Sample Prep Kit v2 (Illumina) was used in the succeeding steps. Generation of mir-7b KO mice. A 15.7-kb fragment encompassing Briefly, total RNA samples (100–1000 ng) were poly-A enriched and mmu-mir-7b was recombined into a minimal vector using homology then reverse transcribed into double-strand cDNA. The cDNA sam- arms-containing primers according to the protocol provided by Gene- ples were fragmented, endrepaired, and polyadenylated before liga- Bridges (http://www.genebridges.com; primers listed in Supplemental tion of TruSeq adapters containing the index sequence for multiplex Table 2). The mir-7b gene was juxtaposed with a loxP-containing neor sequencing. Multiplexing fragments containing TruSeq adapters on cassette 97 bp upstream of the miRNA precursor sequence. Following both ends were selectively enriched with PCR. The quality and quan- expression of the Cre in bacteria, the neor cassette was removed and tity of the enriched libraries were validated using Qubit (1.0) Fluorom- a second Frt/loxP/neor cassette was introduced 498 bp downstream eter and the Caliper GX LabChip GX (Caliper Life Sciences Inc.). The of the miR-7b gene (Supplemental Figure 1A). Upon linearization, the product is a smear with an average fragment size of approximately

jci.org Volume 127 Number 3 March 2017 1071 RESEARCH ARTICLE The Journal of Clinical Investigation

260 bp. The libraries were normalized to 10 nM in Tris-Cl 10 mM, pH mental cage. Mice were reintroduced into the experimental cage, and 8.5, with 0.1% Tween 20. the elapsed time was recorded once they uncovered the chow pellet. Cluster generation and sequencing. The TruSeq PE Cluster Kit Recombinant adenovirus. Mouse miR-7a2–expressing vectors were v3-cBot-HS or TruSeq SR Cluster Kit v3-cBot-HS (Illumina Inc.) was generated by PCR amplification of a 250-bp fragment spanning the used for cluster generation using 10 pM of pooled normalized libraries pre-miR-7a2 genomic sequences. Fragments were cloned at XhoI sites on the cBOT. Sequencing was performed on the Illumina HiSeq 2000 of pAD5 for adenovirus production (Viraquest). The miR-7a adenovi- paired end at 2 × 101 bp or single-end 100 bp using the TruSeq SBS Kit ruses expressed GFP from an independent promoter, and the control v3-HS (Illumina Inc.). adenovirus (Ad-GFP) was identical except that it lacked the miRNA. Data analysis. RNA-Seq reads were quality checked with FastQC Expression was confirmed by TaqMan qPCR. (Babraham Bioinformatics), which computes various quality metrics for Chemical compounds. BMP4, dinoprost, and GnRH were pur- raw reads. Reads were aligned to the genome and transcriptome with chased from Sigma-Aldrich. Grem1 was obtained from R&D Systems. TopHat v 1.3.3 (https://ccb.jhu.edu/software/tophat/manual.shtml). Cell lines. Pituitary cell lines α T3 and L β T2 were provided by P. Before mapping, the low-quality ends of the reads were clipped (3 bas- Mellon (UCD, Davis, California, USA) (58, 59). es from the read start and 10 bases from the read end). TopHat was Antibodies. Antibodies recognizing mouse FSHb and LHb were run with default options. The fragment length parameter was set to obtained from A. Parlow (National Hormone and Peptide Program, 100 bases, with an SD of 100 bases. Based on these alignments, the Harbor-UCLA Medical Center, Torrance, California, USA). TUBB3 distribution of the reads across genomic features was assessed. Iso- (catalog 4466), β-actin (catalog 4970), SMAD1 (catalog 6944), and form expression was quantified with the RSEM algorithm (http://www. phospho-SMAD1/5/9 (catalog 13820) were purchased from Cell Sig- biomedcentral.com/1471-2105/12/323; BMC Bioinformatics) with the naling Technology. CYP17A1 antibody was purchased from Santa Cruz option for estimation of the read start position distribution turned on. Biotechnology Inc. (catalog sc-46081). For seed enrichment analysis, only genes with FPKM above 1.0 were Primer sequences. See Supplemental Table 3. considered. For cumulative distribution function calculations, log2 Statistics. For all experiments, data are shown as mean ± SD if fold-change values were corrected for 3′ UTR length biases (55). not stated otherwise. To determine statistical significance between 2 RNA isolation and real-time pPCR. RNA was isolated using groups, 2-tailed Student’s t test was used. For multiple comparisons, TRIzol reagent (Invitrogen) according to the manufacturer’s proto- 1-way ANOVA followed by Tukey’s post-test was used. A P value of col. RNA was subjected to DNaseI treatment with the DNA-free kit less than 0.05 was considered statistically significant. Statistical anal- (Ambion) when necessary. RNA was reverse transcribed using the ysis was performed using GraphPad Prism 4 (GraphPad software). High Capacity cDNA Reverse Transcription Kit (Applied Biosyste- Study approval. All animal experiments were approved by the Kan- ms). qPCR was performed with a LC480 II Lightcycler (Roche) and tonale Veterinäramt Zürich, Zurich, Switzerland. SYBR Fast Universal Mastermix (Kapa). Results were normalized to Accession codes. All raw RNA-Seq data were deposited in the Euro- 36B4 mRNA levels. miRNA levels were measured using the TaqMan pean Nucleotide Archive (ENA PRJEB12612; http://www.ebi.ac.uk/ena). microRNA Assay Kit (Applied Biosystems), and the results were nor- malized to sno-202 levels. For absolute quantification, synthesized Author contributions mature miRNAs were used (Sigma-Aldrich). KA designed and performed most of the experiments, analyzed Luciferase assays. LbT2 or aT3 cells were cultured in 24-well plates and interpreted data, and wrote the manuscript. ML generated and transfected with 100 ng of pmirGLO reporters. Cells were assayed constitutive and conditional mir-7a1, mir-7a2, and mir-7b KO mice. 48 hours after transfection using the Dual-Luciferase Reporter Assay MPL contributed to in vivo and cell culture experiments and immu- System (Promega). Results were normalized to the Renilla luciferase nohistochemistry. JK performed FSH and LH measurements. RD control contained in pmirGLO and expressed relative to the average performed seed enrichment analysis. EG performed hypothalamus value of the controls. immunohistochemistry. YY contributed to cell culture experiments. GnRH stimulation test. Female nullipara mice were given an intra- TR performed blastocyst injections. MS conceived and supervised peritoneal injection of Buserelin (GnRH , Sigma-Aldrich; 0.01 the project, analyzed and interpreted data, and wrote the manu- μg/μl) in 200 μl of vehicle (PBS) or vehicle alone and blood was col- script. All authors read and commented on the manuscript. lected after 15 minutes for FSH and LH measurements. Hormone measurements. Testosterone and estradiol were quan- Acknowledgments tified using the ELISA Kit from Alpco following the manufacturer’s We wish to thank Hasan Kabakci and Regina Kubsch for excellent instructions. For intratesticular testosterone analysis, lipid extraction technical and animal husbandry assistance and Taina Kirjonen for was performed using 2× diethylether. Serum LH and FSH concentra- skillful assistance with gonadotropin measurements. We are grate- tions were measured using immunofluorometric assays as described ful to Pamela Mellon for providing pituitary cell lines. We thank the previously (55). For measuring intratesticular and serum testosterone, Functional Genomics Center Zurich and the Light Microscopy and both sera and tissue homogenates were extracted twice with 2 ml of Screening Center Zurich for support. This work was supported by diethyl ether, followed by conventional RIA (56). European Molecular Biology Organization (EMBO) long-term fel- Olfaction behavioral test. A buried food test was conducted as pre- lowships (to KA), the Austrian Genome Research Program GEN-AU viously described (57). Nine-week-old mice were fasted for 24 hours II and III (Austromouse) (to TR), European Research Council (ERC) and then were allowed to acclimate in the experimental cage for 5 min- grant “Metabolomirs,” the Starr Foundation International and the utes prior to the test. Mice were transferred to an empty cage, while a Swiss National Science Foundation, and the National Center of Com- 1.5 g pellet of chow was buried 1 cm below the bedding in the experi- petence in Research (NCCR) on RNA Biology and Disease (to MS).

1072 jci.org Volume 127 Number 3 March 2017 The Journal of Clinical Investigation RESEARCH ARTICLE

Address correspondence to: Markus Stoffel, Institute of Molecu- ML’s present address is: Cellular Identity and Metabolism Group, lar Health Sciences, Swiss Federal Institute of Technology (ETH MRC London Institute of Medical Sciences London, United King- Zurich), Otto-Stern Weg 7, 8093 Zurich, Switzerland. Phone: dom; or Institute of Clinical Sciences, Faculty of Medicine, Impe- 41.44.633.4560; E-mail: [email protected]. rial College London, London, United Kingdom.

1. Abreu AP, Kaiser UB. Pubertal development microRNA target site in the GFPT1 gene of 34. Navarro VM, et al. Developmental and hor- and regulation. Lancet Diabetes Endocrinol. patients with congenital myasthenic syndrome. monally regulated messenger ribonucleic acid 2016;4(3):254–264. Hum Mol Genet. 2015;24(12):3418–3426. expression of KiSS-1 and its putative receptor, 2. Herbison AE. Control of puberty onset and fertil- 19. Messina A, et al. A microRNA switch regulates the GPR54, in rat hypothalamus and potent luteiniz- ity by gonadotropin-releasing hormone neurons. rise in hypothalamic GnRH production before ing hormone-releasing activity of KiSS-1 peptide. Nat Rev Endocrinol. 2016;12(8):452–466. puberty. Nat Neurosci. 2016;19(6):835–844. Endocrinology. 2004;145(10):4565–4574. 3. Leon S, Tena-Sempere M. Dissecting the roles of 20. Wang H, et al. Gonadotrope-specific deletion 35. Smith JT, Cunningham MJ, Rissman EF, Clifton gonadotropin-inhibitory hormone in mammals: of Dicer results in severely suppressed gonad- DK, Steiner RA. Regulation of Kiss1 gene expres- studies using pharmacological tools and geneti- otropins and fertility defects. J Biol Chem. sion in the brain of the female mouse. Endocrinol- cally modified mouse models. Front Endocrinol 2015;290(5):2699–2714. ogy. 2005;146(9):3686–3692. (Lausanne). 2016;6:189. 21. Hasuwa H, Ueda J, Ikawa M, Okabe M. miR- 36. Ruzankina Y, et al. Deletion of the developmen- 4. Ramaswamy S, Weinbauer GF. Endocrine control 200b and miR-429 function in mouse ovulation tally essential gene ATR in adult mice leads to of spermatogenesis: Role of FSH and LH/ testos- and are essential for female fertility. Science. age-related phenotypes and stem cell loss. Cell terone. Spermatogenesis. 2015;4(2):e996025. 2013;341(6141):71–73. Stem Cell. 2007;1(1):113–126. 5. Christensen A, et al. Hormonal regulation 22. Latreille M, et al. MicroRNA-7a regulates 37. Michos O, et al. Reduction of BMP4 activity of female reproduction. Horm Metab Res. pancreatic β cell function. J Clin Invest. by gremlin 1 enables ureteric bud outgrowth 2012;44(8):587–591. 2014;124(6):2722–2735. and GDNF/WNT11 feedback signalling during 6. Brioude F, et al. Non-syndromic congenital hypo- 23. Choudhury NR, et al. Tissue-specific control of branching morphogenesis. Development. gonadotropic hypogonadism: clinical presenta- brain-enriched miR-7 biogenesis. Genes Dev. 2007;134(13):2397–2405. tion and genotype-phenotype relationships. Eur J 2013;27(1):24–38. 38. Orlicky DJ. Negative regulatory activity of a pros- Endocrinol. 2010;162(5):835–851. 24. Memczak S, et al. Circular RNAs are a large class taglandin F2 alpha receptor associated protein 7. Stamou MI, Cox KH, Crowley WF Jr. Discovering of animal RNAs with regulatory potency. Nature. (FPRP). Prostaglandins Leukot Essent Fatty Acids. genes essential to the hypothalamic regulation 2013;495(7441):333–338. 1996;54(4):247–259. of human reproduction using a human disease 25. Hansen TB, et al. Natural RNA circles func- 39. Orlicky DJ, Lieber JG, Morin CL, Evans RM. model: adjusting to life in the -omics Era. Endocr tion as efficient microRNA sponges. Nature. Synthesis and accumulation of a receptor Rev. 2015;36(6):603–621. 2013;495(7441):384–388. regulatory protein associated with lipid drop- 8. de Roux N, Carel JC, Léger J. Congenital hypo- 26. Horsham JL, Ganda C, Kalinowski FC, Brown let accumulation in 3T3-L1 cells. J Lipid Res. gonadotropic hypogonadism: a trait shared by RA, Epis MR, Leedman PJ. MicroRNA-7: A 1998;39(6):1152–1161. several complex neurodevelopmental disorders. miRNA with expanding roles in develop- 40. Ahn J, Febbraio M, Silverstein RL. A novel Endocr Dev. 2016;29:72–86. ment and disease. Int J Biochem Cell Biol. isoform of human Golgi complex-localized gly- 9. Topaloglu AK, Kotan LD. Genetics of hypogonado- 2015;69:215–224. coprotein-1 (also known as E-selectin ligand-1, tropic hypogonadism. Endocr Dev. 2016;29:36–49. 27. Gu DN, Huang Q, Tian L. The molecular MG-160 and cysteine-rich fibroblast growth 10. Margolin DH, et al. Ataxia, dementia, and hypo- mechanisms and therapeutic potential of factor receptor) targets differential subcellular gonadotropism caused by disordered ubiquitina- microRNA-7 in cancer. Expert Opin Ther Targets. localization. J Cell Sci. 2005;118(Pt 8):1725–1731. tion. N Engl J Med. 2013;368(21):1992–2003. 2015;19(3):415–426. 41. Exton JH. Regulation of phosphoinositide phos- 11. Boehm U, et al. Expert consensus document: 28. Junn E, Lee KW, Jeong BS, Chan TW, Im JY, pholipases by hormones, neurotransmitters, and European Consensus Statement on congenital Mouradian MM. Repression of alpha-synuclein other linked to G proteins. Annu Rev hypogonadotropic hypogonadism--pathogenesis, expression and toxicity by microRNA-7. Proc Natl Pharmacol Toxicol. 1996;36:481–509. diagnosis and treatment. Nat Rev Endocrinol. Acad Sci USA. 2009;106(31):13052–13057. 42. Nicol L, Faure MO, McNeilly JR, Fontaine J, 2015;11(9):547–564. 29. Amador A, Parkening T, Beamer W, Bartke A, Taragnat C, McNeilly AS. Bone morphogenetic 12. Guo H, Ingolia NT, Weissman JS, Bartel DP. Collins TJ. Autoregulation of testicular lutein- protein-4 interacts with activin and GnRH to Mammalian microRNAs predominantly izing hormone receptors in hypogonadal (hpg/ modulate gonadotrophin secretion in LbetaT2 act to decrease target mRNA levels. Nature. hpg) mice. Biochem Genet. 1984;22(5-6):395–401. gonadotrophs. J Endocrinol. 2008;196(3):497–507. 2010;466(7308):835–840. 30. Good DJ, Porter FD, Mahon KA, Parlow AF, West- 43. Faure MO, et al. BMP-4 inhibits follicle- 13. Friedman RC, Farh KK, Burge CB, Bartel DP. phal H, Kirsch IR. Hypogonadism and obesity in stimulating hormone secretion in ewe pitu- Most mammalian mRNAs are conserved targets mice with a targeted deletion of the Nhlh2 gene. itary. J Endocrinol. 2005;186(1):109–121. of microRNAs. Genome Res. 2009;19(1):92–105. Nat Genet. 1997;15(4):397–401. 44. Grattan DR. 60 Years of neuroendocrinology: 14. Landgraf P, et al. A mammalian microRNA 31. Eldar-Geva T, Hirsch HJ, Benarroch F, Rubinstein The hypothalamo-prolactin axis. J Endocrinol. expression atlas based on small RNA library O, Gross-Tsur V. Hypogonadism in females with 2015;226(2):T101–T122. sequencing. Cell. 2007;129(7):1401–1414. Prader-Willi syndrome from infancy to adult- 45. Tchernof A, Després JP. Sex steroid hormones, sex 15. Mendell JT, Olson EN. MicroRNAs in hood: variable combinations of a primary gonad- hormone-binding globulin, and obesity in men and stress signaling and human disease. Cell. al defect and hypothalamic dysfunction. Eur J women. Horm Metab Res. 2000;32(11-12):526–536. 2012;148(6):1172–1187. Endocrinol. 2010;162(2):377–384. 46. Sanek NA, Young WS. Investigating the in vivo 16. de Pontual L, et al. Germline deletion of the miR- 32. Denzler R, Agarwal V, Stefano J, Bartel DP, expression patterns of miR-7 microRNA family 17~92 cluster causes skeletal and growth defects Stoffel M. Assessing the ceRNA hypothesis with members in the adult mouse brain. Microrna. in humans. Nat Genet. 2011;43(10):1026–1030. quantitative measurements of miRNA and target 2012;1(1):11–18. 17. Mencía A, et al. Mutations in the seed region abundance. Mol Cell. 2014;54(5):766–776. 47. Kim SH. Congenital Hypogonadotropic Hypo- of human miR-96 are responsible for nonsyn- 33. Horseman ND, et al. Defective mammopoiesis, gonadism and Kallmann Syndrome: Past, dromic progressive hearing loss. Nat Genet. but normal hematopoiesis, in mice with a tar- Present, and Future. Endocrinol Metab (Seoul). 2009;41(5):609–613. geted disruption of the prolactin gene. EMBO J. 2015;30(4):456–466. 18. Dusl M, et al. A 3’-UTR mutation creates a 1997;16(23):6926–6935. 48. Young LS, Speight A, Charlton HM, Clayton

jci.org Volume 127 Number 3 March 2017 1073 RESEARCH ARTICLE The Journal of Clinical Investigation

RN. Pituitary gonadotropin-releasing hormone blood flow in mares. Domest Anim Endocrinol. and luteinizing hormone and application on sera of receptor regulation in the hypogonadotrophic 2007;32(4):315–328. cycling rats. Biol Reprod. 2000;62(4):886–894. hypogonadal (hpg) mouse. Endocrinology. 53. Massagué J, Blain SW, Lo RS. TGFbeta signaling 57. Yang M, Crawley JN. Simple behavioral assess- 1983;113(1):55–61. in growth control, cancer, and heritable disor- ment of mouse olfaction. Curr Protoc Neurosci. 49. Behrman HR. Prostaglandins in hypothalamo- ders. Cell. 2000;103(2):295–309. 2009;Chapter 8:Unit 8.24. pituitary and ovarian function. Annu Rev Physiol. 54. Paez-Pereda M, et al. Involvement of bone 58. Windle JJ, Weiner RI, Mellon PL. Cell lines of the 1979;41:685–700. morphogenetic protein 4 (BMP-4) in pituitary pituitary gonadotrope lineage derived by target- 50. Labhsetwar AP. Effects of prostaglandin prolactinoma pathogenesis through a Smad/ ed oncogenesis in transgenic mice. Mol Endocri- F2-alpha on pituitary luteinizing hormone estrogen receptor crosstalk. Proc Natl Acad Sci nol. 1990;4(4):597–603. content of pregnant rats: a possible explana- USA. 2003;100(3):1034–1039. 59. Thomas P, Mellon PL, Turgeon J, Waring DW. tion for the luteolytic effect. J Reprod Fertil. 55. Haavisto AM, Pettersson K, Bergendahl M, Per- The L beta T2 clonal gonadotrope: a model for 1970;23(1):155–159. heentupa A, Roser JF, Huhtaniemi I. single cell studies of endocrine cell secretion. 51. Labhsetwar AP. Do prostaglandins stimulate LH A supersensitive immunofluorometric assay Endocrinology. 1996;137(7):2979–2989. release and thereby cause luteolysis? Prostaglan- for rat luteinizing hormone. Endocrinology. 60. Garcia DM, Baek D, Shin C, Bell GW, Grimson A, dins. 1973;3(5):729–732. 1993;132(4):1687–1691. Bartel DP. Weak seed-pairing stability and high 52. Ginther OJ, Gastal EL, Gastal MO, Beg MA. 56. van Casteren JI, Schoonen WG, Kloosterboer HJ. target-site abundance decrease the proficiency of Effect of prostaglandin F2alpha on ovarian, Development of time-resolved immunofluoro- lsy-6 and other microRNAs. Nat Struct Mol Biol. adrenal, and pituitary hormones and on luteal metric assays for rat follicle-stimulating hormone 2011;18(10):1139–1146.

1074 jci.org Volume 127 Number 3 March 2017